Empower your writing process


Authors. Caroline Salamin, Noémi Cobolet, Pascale Bouton, Raphaël Grolimund, Mathilde Panes (Bibliothèque de l’EPFL)

Document type. This document is not a scientific paper, but the course notes of a seminar for PhD students provided by the EPFL Library. This seminar is module 2.

Version. This is v1.0.8 of the document, updated on May 15, 2018. The latest version of this document is available at http://go.epfl.ch/phd-module2.

Copyright. CC BY-NC-SA Bibliothèque de l’EPFL | This document is distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License.

Additional material. Two appendices are handed out at the end of the session: a general appendix (https://www.authorea.com/users/26260/articles/61573/) containing examples and answers; and an appendix specific to the session, created by the participants. The session-specific appendices are not published online.

Acknowledgments. The authors would like to thank Vijay Kartik who accepted to take part in the seminar as guest speaker and enrich it with his experience as former PhD students at EPFL. His presentation is available here.
The authors also want to thank Vijay Kartik for fixing some minor errors in this document.

Abstract. This document is tailored for PhD Students interested in facilitating their writing process. We look at the strategies to improve the findability of your papers by taking advantage of the mechanisms provided by major scientific databases. We then show how to avoid pitfalls of collaborative writing (authorship, versioning and collaborative tools). We finally explain how to properly reuse other works in your publications. The in-class session also includes practical exercises in Authorea.

Boost your paper findability

Introduction: what are we talking about?

This section will focus on the findability of scientific articles that are meant to be published in peer-reviewed journals.

To be visible, your paper must be disseminated widely, easy to find and access, and frequently cited by other scholars. Concerning findability, there are 2 essential steps to take care of. First of all, it is of vital importance to understand the mechanisms of the main search engines and to keep them in mind to prepare your papers accordingly! Take time to explore the resources where your paper might be listed. To do so, you may check the articles published by journals in your domain and observe where and how they are indexed. When this is clear to you, build your title and abstract, and keywords with great care, following essential rules that we will look over.

Search engines and relevance

Authors need to be conscious of the indexing mechanisms of scientific search engines and bibliographic databases: then they will be able to use them in drawing up the paper.

Relevance sort is usually based on algorithms that take into account :

  • the number of occurrences (times that the searched words appear in the records)

  • whether the words are found as an exact phrase or separately

  • when words are found separately, closer proximity ranks higher

  • whether the words are found within metadata fields designated as particularly relevant (title field...)

  • the words’ location within the field or within the full text document.

Beside these basic elements, search engines and databases add their specific criteria.

see details about Google Scholar and Web of Science in appendix:


Make the most of your title, abstract and keywords

The title, abstract and keywords should be chosen with great care, since they will facilitate computer-based search and retrieval (Hartley 2008)

For every person who reads the whole of a scientific paper, about 500 read only the title (Kerkut 1983, in (Gustavii 2008)). This sentence points out how choosing the right words for your title is crucial. It will help the reader decide if he wants to read the full article or not in just a few seconds. It should therefore clearly indicate precise information on the content of the article. An article with a badly chosen title may virtually disappear and never be found. Below are a few tips to keep in mind when you build your title:

  • Opt for a title that suggests you are committed, rather than a neutral one.

  • Hanging titles imply the use of punctuation, more words, so more risk of including waste words... not a good idea.

  • Use a verb to make a sentence, which will have stronger impact than just a phrase. Moreover, verbs bring dynamism to the title.

  • Avoid titles formulated as questions. The answer will be more accurate to choose an article when scanning a list of titles.

  • Include specific rather than general terms. Imagine the keywords a researcher would use to find your article.Think of the significant content of the article and extract the keywords you will use for your title. Place these keywords at the beginning of the title to make it immediately clear what you are talking about.

  • Use the fewest words, keep only the descriptive ones, and avoid the waste words (Day 2006) like “an observation on”, “investigations on”, “a”, “the”, “is”... which do not give any information on the content. Short titles are more comprehensible than long ones.

  • Be careful with syntax! Grammatical errors in titles are a common pitfall easy to avoid.

  • Avoid abbreviations unless the abbreviation is better known than the full name (e.g. DNA).

  • Avoid jargon

Writing a well-structured and striking abstract is also essential since it will help the reader decide quickly if reading the full content is useful for his research. Not to forget that the abstract is the first part reviewers read and that it can thus determine if your paper will be accepted by a journal. What’s more, an abstract is requested to be a speaker in conferences. So, we can say that it is an art to master!

The abstract can be seen as the table of contents of an article. Most of the time, an abstract consists of 4 basic sections: background/introduction, materials and methods, results and discussion/conclusion. Journals usually give instructions on the use or not of explicit headlines. To ensure your abstract is well-structured, keep these 4 sections in mind while building it even if the subtitles are not required.

Below are a few tips to keep in mind when you build your abstract:

  • Abstracts are often published alone (in Web of Science, for example), so do not make any reference to the full text like “as shown on figure 2”. It should therefore be understandable by itself.

  • Like in the title, do not use abbreviations, unless they are better known than the full name.

  • Eliminate verbiage. If you can use up to 200 words, and consider that 150 are enough, just keep it shorter.

Last but not least, choose author keywords carefully, using specific terms if you plan to publish in a specialized journal.

Summary of the section

Collaborative writing : method, tools

Today, researchers hardly publish a paper alone. Fig. \ref{fig:authorsperpaper} shows that the average number of authors per paper has increased from 2 or less in 1950 up to 5 in 2005.

\label{fig:authorsperpaper}Evolution of the average number of authors per paper (reproduced from Fig. 2C published under CC BY license in (Wallace 2012))

Writing collaboratively has big advantages: a greater knowledge and skills base, a better understanding of the audience…, but also substantial drawbacks: it may take more time, you can face disjointed style or content, work load could be inequitably distributed, smaller role can reduce motivation to work, conflicts or ill-will can arise…

Remember that the preparatory-phase is crucial !

Then you will have to:

  • discuss and agree on the demonstration aims and on the argumentation,

  • express them together in a detailed plan of the paper to be produced,

  • determine its style and length (according to editors’ guidelines),

  • allocate roles between people : who will be author ? who will be supervisor ? who will be a proofreader ?

  • according to these roles, distribute the different parts of the paper or chapters between authors,

  • enter the table of contents, with an author name across from each chapter,

  • fix a precise timetable, with different milestones : writing phase, proofreading phase, revision phase, second proofreading phase... and deliverables.

\label{fig:authorship}Conditions for being credited as author (CC BY Bibliothèque de l’EPFL)


In order to be considered as an author, a researcher must fulfill the 3 following criteria (see Fig. \ref{fig:authorship}):

  1. have made an essential contribution to the planning, carrying out, evaluation and verification of the research work;

  2. have participated in the writing of the manuscript;

  3. and have approved the final version of the manuscript.

What to do when someone gave an important input without fulfilling all the criteria to be considered as an author? Other persons who have contributed to the study, but only partially fulfill the criteria [...], must be acknowledged [...], but are not designated as authors. (EPFL 2013, Brueckner 2013)

And, in the list of authors, who will appear first or last? There is not one easy answer, but this complex question has to be answered by all the authors as quickly as possible in the project.

It’s also important to mention that “[a]ll persons fulfilling the criteria for authorship must be listed as authors of a scientific publication.” (Brueckner 2013) Someone who doesn’t appear as author of a publication he significantly contributed to is called a ghostwriter. On famous and funny example is the paper, still known as the alpha-beta-gamma paper (Alpher 1948), because the authors, Ralph Alpher and George Gamow, added the name of one of their friend, Hans Bethe, so that the authors list looks like Alpher, Bethe, Gamow. One last detail: the paper was published on April 1st, 1948...

On the contrary, granting authorship to someone who doesn’t deserve it (based on the criteria listed above) is called honorary authorship (or gift authorship). Guest authorship is similar to honorary authorship, but there is an additional “expectation that [the] inclusion of a particular name will improve the chances that the study will be published”. (CSE 2012) Additionally, a paper should never be anonymous like (Student 1908) or (Anon. 1970).

All of these practices are violations of scientific integrity (see Fig. \ref{fig:GAGG}). There are rare cases (e.g. when the author’s life is threatened) where the editor can decide to publish a paper anonymously or to use a pseudonym.

Finally, for large research groups, publications can be credited to the group. This is called group authorship. The world record of the number of authors for a paper is 5154! As you can see this article (http://dx.doi.org/10.1103/PhysRevLett.114.191803), the 2 research groups (ATLAS Collaboration & CMS Collaboration) are mentioned as authors. But you have the complete list of the authors at the end of the paper.

\label{fig:GAGG}GAGG Authorship: inappropriate facets of authorship (CC BY Bibliothèque de l’EPFL)

Collaborative writing technical issues

When working on a paper with co-authors, a few questions raise:

  • who has access to the latest version of the paper?

  • who has access to the figures included in the paper?

  • who has access to the references cited in the paper?

These problems become a nightmare if you use a word processor like Microsoft Word or LibreOffice Writer. But the two first problems seem to be easy to fix: use a Dropbox-like system (or any shared location, if all co-authors work in the same institution). But it doesn’t work for the third one. To share the references, you need a group library or to integrate them in the document in a way that lets co-authors manipulate them. (We’ll see how to do that.) Versioning a paper also allows you to restore a previous version if needed.

When thinking about authorship, versioning can sometimes also answer a tricky question: who wrote what? (see Fig. \ref{fig:dynamicpublications})

Mastering versioning of scientific papers could lead to what (Heller 2014) call dynamic publication formats. Instead of writing a brand new paper when research results have been improved or completed, the original paper could be updated.

Collaborative writing

Bosman and Kramer (Bosman; 2015) identified 6 phases in the research workflow and described the tools and sites related to them. In the writing phase, the infographic shows that before 2010 no tool for collaborative writing was available apart from Google Docs. Nowadays, some online tools appeared: the most important of them are Authorea, OverLeaf and ShareLaTeX.

\label{fig:101innovativetools}101 innovative tools and sites in 6 research workflow phases (<2000-2015) source: (Bosman; 2015)

Authorea, OverLeaf and ShareLaTeX provide similar features: they produce LaTeX documents without any need for mastering LaTeX (thanks to a graphic user interface), make collaborative writing easy and provide document history (versioning with the possibility to restore an older version of the document). Depending on the tool, you can also rely on spellchecking, change tracking or take advantage of publisher style layouts and offline editing. The main difference between these tools is summarized in Fig. \ref{fig:AOS}. As EPFL pays a site license for Authorea, all members of the EPFL community can have unlimited private documents.

\label{fig:AOS}Online collaborative writing tools: difference between Authorea, OverLeaf and ShareLaTeX

These tools enable several authors to easily write an article together by always providing access to the latest version of the paper (including bibliographic references and figures). But these tools could be used to create dynamic publication formats. Fig. \ref{fig:todayVSdynamic} illustrates what this means: publications on a very same topic could be combined instead of being multiplied. The idea is not to publish the same results again and again, nor to publish each new small bit of new results. Peer reviewers are meant to filter out every submitted paper with no real novelty or originality.

\label{fig:todayVSdynamic}Comparison of today’s dynamic publication and dynamic publication (reproduced from (Heller 2014), fig. 1 CC BY-NC)

Another novelty in dynamic publication is the ease to attribute contributions to their authors (at least those who inserted them in the tool). But working this way makes it easy for the editor or the reviewer to make sure a researcher who didn’t contribute to the writing of the paper doesn’t put his name on it.

There are many open questions left, but the concept of dynamic scholar publication and new online collaborative writing tools allow researchers to try to fix some problems that scientific publication is facing.

Citation good practices

Another big challenge in scientific publication is related to citation. There are 3 questions to answer to be sure to cite properly:

Why should I cite? What should I cite? How should I cite?

Why should I cite?

This answer to this question seems obvious: scientific ethics requires an author to acknowledge someone else’s work when his article is based on it. But there is more.

You write a paper to share your results with the scientific community. Ask yourself who do I write for? The answer is “the reader”. Think about the reader’s needs. It should help you when you ask yourself “what should I cite?” and “How should I cite?”.

What should I cite? (And what should not be cited?)

Here are the general rules of what you should cite. Examples and counterexamples are provided in the appendix.

Always (read and) cite the original document and don’t rely on a paper that cites a previous one. The author you read could have made mistakes. There is one notable exception: when the language of the original document is not intelligible to the targeted audience.

Try to be as specific as possible! You wouldn’t cite a whole journal, you cite a specific paper. Do the same when possible (e.g. books or websites)! Help the reader to find and access your sources quickly.

Don’t cite an unpublished paper! First, the reader doesn’t have access to it. How can he/she be sure that you didn’t make all this up. Second, the paper can be rejected or published with major corrections. What you cite could disappear from the final version of the paper!

Don’t cite informal information! Again, the reader can’t access what someone told you. If you cite it, it’s useless for the reader. Thank the provider of the information in the acknowledgments.

How should I cite?

A bibliographic reference is made of metadata. You can add more information, but it must at least contain all information needed for the reader to find it. (l’EPFL 2015) explains what the minimal pieces of information are for the commons document types. You are very welcome to add as many useful information as you can to help the reader.

Be as explicit as possible! Most of the time, you can’t choose the citation style: the publisher requires a specific layout for the in-text citations and the bibliography. But if you have the choice, be aware that an in-text citation like [1] doesn’t provide any information about the author or the publication date. You have this information with (Heller 2014).

Again, be as specific as possible! If possible, add the page number in the in-text citation when quoting. (Heller 2014) leads to the reference in the bibliography. (Heller 2014, p. 202) also leads to the right page in that document.

As already mentioned, most of the time you don’t choose the citation style. But be aware that not all citation styles are equal in terms of information provided. A few provide full names whereas others only give initials for first names. A few abbreviate journal names, others don’t. A few provide article titles, others don’t! (see Appendix)

Good citation practices are meant to help the reader. This requires giving as much information as possible. The author’s work is to make sure that the references are complete and correct. Then a software formats the elements (and sometimes cuts some of them out) according to the citation style of the targeted journal.

Illustration of your publications

First part of the game "Copyright compliance lotto"

Second part of the game “Copyright compliance lotto”

You should remember the main principles below to make fair use of images and graphs. Images and graphs are considered as integral works, so you cannot use part of them, like you may do for text articles. You always use the integral work, which means that citing is essential, but not enough. You also need to either:

  • ask the author for permission (and publisher if applicable)

  • or find an image/graph under a Creative Commons license

  • or build your own graph or create your own image. The data are not protected, only the format is.

The above rules apply when your work is meant to be published.

If you only need to use the image/graph for teaching purposes (i.e. online access is restricted to your students), you need to cite your sources of course, but you do not need to ask for an authorization.

If you participate in a MOOC (course online, free access) or if you share your course and slides on the web, you are not covered by “teaching purposes” anymore.

If the image/graph is for internal use only (i.e. for your laboratory): you need to cite your sources of course, but you do not need to ask for an authorization.

For more information, please consult this page: http://citation.epfl.ch/copyright/reuse-works.


The reference list below is a selection of the complete version of the bibliography that stands and evolves online: https://www.zotero.org/groups/phd-seminar/items/collectionKey/PXVSMEV3


  1. James Hartley. Academic writing and publishing : a practical handbook. Routledge, 2008.

  2. Björn Gustavii. How to write & illustrate a scientific paper. Cambridge University Press, 2008.

  3. Robert Adams Day, Barbara Gastel. How to write and publish a scientific paper. Cambridge University Press, 2006.

  4. Matthew L. Wallace, Vincent Larivière, Yves Gingras. A Small World of Citations? The Influence of Collaboration Networks on Citation Practices. PLoS ONE 7, e33339 (2012). Link

  5. EPFL. Directive concerning research integrity and good scientific practice at EPFL. (2013). Link

  6. Christian Brueckner, Sybille Ackermann, Michelle Salathé. Authorship in scientific publications: analysis and recommendations. Swiss academies of arts and sciences, 2013. Link

  7. R. A. Alpher, H. Bethe, G. Gamow. The Origin of Chemical Elements. Physical Review 73, 803–804 American Physical Society (APS), 1948. Link

  8. CSE. CSE’s White Paper on Promoting Integrity in Scientific Journal Publications. Council of Science Editors, 2012. Link

  9. Student. The probable error of a mean. Biometrika 6, 1–25 Oxford University Press (OUP), 1908. Link

  10. Anon.. Effects of Sexual Activity on Beard Growth in Man. Nature 226, 869–870 Springer Nature, 1970. Link

  11. Lambert Heller, Ronald The, Sönke Bartling. Dynamic Publication Formats and Collaborative Authoring. 191–211 In Opening Science. Springer International Publishing, 2014. Link

  12. Bianca Kramer; Jeroen Bosman;. 101 Innovations in Scholarly Communication - the Changing Research Workflow. Figshare, 2015. Link

  13. Bibliothèque de l’EPFL. Rational Bibliographic: guide de rédaction des références bibliographiques. EPFL, 2015. Link

[Someone else is editing this]

You are editing this file